Atlantic—Automated data preprocessing framework for supervised machine learning
Luís Santos, Luís Ferreira · Software Impacts · 2023
Atlantic is an open-source Python package designed to simplify and automate data preprocessing for supervised ML (Machine Learning) tasks. The Atlantic package integrates multiple customizable preprocessing mechanisms, including datetime feature engineering, automated feature selection, categorical encoding techniques and null imputation methods. In order to provide a comprehensive approach to data processing automation, Atlantic automated preprocessing pipeline follows an optimization method based on tree-based models ensembles. The main goal of Atlantic is to automatically identify the best combination of preprocessing mechanisms for a specific dataset, aiming to improve the performance of future ML models applied to that data.